[
https://issues.apache.org/jira/browse/HDDS-7321?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17685698#comment-17685698
]
Sammi Chen commented on HDDS-7321:
----------------------------------
[~sadanand_shenoy] The level threshold limit for auto-compaction is usually
high, say 1GB. Imported sst file of a container is generally very small, say
several KB. So there could be thousands or even more small sst files, and still
far from the auto-compaction threshold, that's why we need this auto small sst
file compaction. Two many small sst files will reduce rocksdb's performance
and stability.
> Auto rocksDB small sst files compaction
> ---------------------------------------
>
> Key: HDDS-7321
> URL: https://issues.apache.org/jira/browse/HDDS-7321
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Sammi Chen
> Assignee: Sammi Chen
> Priority: Major
> Labels: pull-request-available
> Fix For: 1.4.0
>
>
> RocksDB has auto compaction itself, which is triggered by the total level
> file size.
> Once the total level file size exceeds the threshold, RocksDB will schedule
> the compaction in backend.
>
> When replicating containers between datanodes, current implementation
> leverages RocksDB SstFileWriter to export container meta data to individual
> sst files, and leverages RocksDB ingestExternalFile to import container meta
> data sst files directly into target datanode RocksDB. If the imported
> container meta data keys don't overlap with other sst files(Consider Merge
> RocksDB design, container ID is used as prefix of each meta data key, this
> is true for most of time), the imported sst file will be kept and remain the
> same without compacting with other existing sst files.
> The worst case, if thousands or dozens of thousands of containers are
> imported on one datanode, there would be dozens of thousands of small sst
> files under one RocksDB, or across all RocksDB instances of one datanode. By
> default, RocksDB has no limit of open files. Dozens of thousands of small
> sst files would exhaust the process open file quota, and service stability
> will be impacted.
> This task aims to provide a way to auto compact all these small sst files
> into merged big ones. Of course, the compaction will have impact on user
> data read/write performance on this datanode.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
